Takeaways from ACT-IAC’s Federal Insights Exchange with USDA’s Fredy Diaz

ACT-IAC Federal Insights Exchange banner

Written by:

Mar 20, 2024

ACT-IAC Federal Insights Exchange

The American Council for Technology and Industry Advisory Council (ACT-IAC) created the Federal Insights Exchange (FIE) program to serve as a dynamic platform for fostering collaboration, innovation, and knowledge exchange within the federal government. The FIE program is underwritten by CGI Federal, HumanTouch, and LMI.

ACT-IAC hosted more than 80 people on Tuesday, March 19, for a virtual Federal Insights Exchange (FIE) session featuring Fredy Diaz, the Deputy Chief Data Officer at the U.S. Department of Agriculture (USDA).

Diaz’s described the USDA’s recent data journey and enterprise approach to transform into a data-driven decision-making organization. He also talked about the advent of the Chief Artificial Intelligence Officer (CAIO) role with the Office of the CDO and how the USDA is creating its first AI strategy.

Included below are 6 key takeaways from Diaz’s presentation. (The event recording is available on the ACT-IAC website for a $15 fee.) This FIE session was moderated by Douglas Bailey, principal at Deep Water Point & Associates, who is the former Chief Technology Officer (Acting) at the USDA and former CIO of USDA’s Agricultural Marketing Service mission area.


6 Key Takeaways from the ACT-IAC FIE

  1. Human Capital in data and AI is as important as anything to do with strategy, technology, or governance.
    The USDA is working on upskilling, recruiting, and retaining a data workforce. USDA has a Data Science training program that emerged organically from the Food and Nutrition Service and has been scaled up to be the way they upskill data workers. They have recently started a cohort of Data and AI interns who are embedded in the eight Mission Areas. And Diaz says they need to make USDA a desirable place to go to work and stay, so they can compete with other federal agencies and the private sector. “We’re always looking for new innovative and unorthodox ideas in this space,” Diaz says. “I think we’re pushing the envelope and beginning to – to be completely honest – make people uncomfortable, which is good, because we need to be able to compete. We need to be able to retain the best talent. And that means pushing our back-office functions a little bit. It means asking questions that others may not have asked in the past.”
  2. Industry partners need to align their solutions with the USDA Data Strategy and understand current strategic priorities.
    When asked for guidance to industry about how to engage with the USDA OCDO, Diaz said that companies should read the 2024-2026 Data Strategy (PDF), understand the needs of the CDO and the Mission Areas, and align their solutions with the strategic objectives. This shows Department leaders that the partner has done their market research – just like the Department is required to do its own market research on available solutions. “Sometimes I get sold a faster car, but really I need the roads and the better-paved roads” to get to the destination. And he says vendors should have patience: “Sometimes we may not be ready to fully embrace it, but it’s not that we’re not interested. We may have to lay some foundational work ahead of time through convincing or funding.”
  3. The USDA EDAPT platform will both shrink and grow in the next few years.
    The USDA’s Office of the CDO has full control over the tools and technologies that comprise its Enterprise Data Analytics Platform and Toolset (EDAPT), which is somewhat unusual among federal agencies. They bear the full responsibility for ATO, certifications, and standards, but it also gives them flexibility and control over what they can and cannot do. Reporting to the CIO and having a strong peer relationship with the CTO makes this a little easier. Diaz says a priority under the new Data Strategy is to listen to feedback from the eight Mission Areas about EDAPT so they can decide what tools to sunset and which to double-down on based on demand. They will modernize the platform and rationalize the choices from previous years. Diaz says they will likely create common, shared platforms as enterprise services, which will allow them to consolidate servers and save on licenses where several Mission Areas have the same platforms.
  4. The USDA is way ahead with AI and is going slow with generative AI.
    The USDA has already made public nearly 40 AI use cases, as required by Section 5 of Executive Order (EO) 13960, “Promoting the Use of Trustworthy Artificial Intelligence in the Federal Government.” The Department has used various AI technologies for several years – whether it’s Natural Language Processing, deep learning, or supervised and unsupervised Machine Learning. But they are taking a more cautious, experimental approach with generative AI. USDA’s CDO, Chris Alvares, is now also the Chief AI Officer. So they have collaborated with the CTO to stand up a GenAI innovation lab to capture new use cases. “We’re not looking to move fast in this. We’re looking to move far. This means setting up the infrastructure, talking with senior leaders, and really getting everyone to understand what it is. So we’re not just running to do proofs of concept.” Diaz says they have focused so far on the big-picture, enterprise perspective of generative AI.
  5. USDA will publish an AI Strategy in the next few months.
    Diaz said the CDO and CAIO teams are working on the AI Strategy – assessing the current state, doing surveys, asking questions, and watching interviews. “Expect it to not just be technology choices or use cases, but rather, what are we doing with the workforce? How are we preparing the workforce to embrace artificial intelligence? How are we preparing our governance and our leadership to understand both the good and the bad of artificial intelligence?” Diaz also says it will be an on-going effort to align the AI Strategy not only with the USDA Data Strategy but also the Department’s IT Strategic Plan and the USDA’s overall business strategy.
  6. Building internal partnerships leads to increased trust and data sharing.
    The barriers to data sharing internally are mostly cultural, not technical. Data owners often ask “Why should I share this with others?” and look at data as a tactical asset, rather than a strategic asset, which is the mindset that Diaz and others want to foster. Through the USDA’s identity system linkages with the EDAPT platform, data owners have control to share it with a finite number of people or remain in the approval chain for access. “Data sharing is actually a form of trust. As trust increases, data sharing increases,” he said.


You May Also Like…


Skip to content